面向对象的高分辨率遥感影像土地覆盖信息提取

利用高分辨率影象提取土地覆盖信息的关键技术在于如何利用丰富的纹理信息来弥补光谱信息的不足。面向对象的图像分类技术改变了传统的面向像素的分类技术:(1)用来解译图像的信息并不在单个像元中,而是在图像对象和其相互关系中;采用多分辨率对象分割方法生成图像对象,提高了分类信息的信噪比;基于对象的分类技术不同于纯粹的光谱信息分类,图像对象还包含了许多的可用于分类的一些其他特征:形状、纹理、相互关系、上下关系等信息。面向对象的土地覆盖分类结果与传统分类方法相比,其特征提取算子更加地适合于几何信息和结构信息丰富的高分辨率图像的自动识别分类。

Object-oriented land cover information extraction from high-resolution remote sensing images

Using high-resolution images from land cover information technology is the key to how to make use of rich texture spec- trum of information to make up for the lack of information.Object-oriented image classification technology has changed the traditional classification of the pixel-oriented technology:To interpret the image information is not in a single pixel,but in the image objects and their mutual relations;using multi-resolution image segmentation method of generating targets,raising the classification of information SNR;Object-based classification is different from a purely technical spectrum of information classification,image also contains many objects can be used for the classification of some other characteristics:shape,texture,mutual relations,relations and context etc. Based on eCognition land cover classification results compared with the traditional classification,feature extraction and operator for more information on the geometry and structure of information-rich high-resolution images of automatic identification classification.